Artificial Intelligence Nanodegree

Convolutional Neural Networks

Project: Write an Algorithm for a Dog Identification App


In this notebook, some template code has already been provided for you, and you will need to implement additional functionality to successfully complete this project. You will not need to modify the included code beyond what is requested. Sections that begin with '(IMPLEMENTATION)' in the header indicate that the following block of code will require additional functionality which you must provide. Instructions will be provided for each section, and the specifics of the implementation are marked in the code block with a 'TODO' statement. Please be sure to read the instructions carefully!

Note: Once you have completed all of the code implementations, you need to finalize your work by exporting the iPython Notebook as an HTML document. Before exporting the notebook to html, all of the code cells need to have been run so that reviewers can see the final implementation and output. You can then export the notebook by using the menu above and navigating to \n", "File -> Download as -> HTML (.html). Include the finished document along with this notebook as your submission.

In addition to implementing code, there will be questions that you must answer which relate to the project and your implementation. Each section where you will answer a question is preceded by a 'Question X' header. Carefully read each question and provide thorough answers in the following text boxes that begin with 'Answer:'. Your project submission will be evaluated based on your answers to each of the questions and the implementation you provide.

Note: Code and Markdown cells can be executed using the Shift + Enter keyboard shortcut. Markdown cells can be edited by double-clicking the cell to enter edit mode.

The rubric contains optional "Stand Out Suggestions" for enhancing the project beyond the minimum requirements. If you decide to pursue the "Stand Out Suggestions", you should include the code in this IPython notebook.


Why We're Here

In this notebook, you will make the first steps towards developing an algorithm that could be used as part of a mobile or web app. At the end of this project, your code will accept any user-supplied image as input. If a dog is detected in the image, it will provide an estimate of the dog's breed. If a human is detected, it will provide an estimate of the dog breed that is most resembling. The image below displays potential sample output of your finished project (... but we expect that each student's algorithm will behave differently!).

Sample Dog Output

In this real-world setting, you will need to piece together a series of models to perform different tasks; for instance, the algorithm that detects humans in an image will be different from the CNN that infers dog breed. There are many points of possible failure, and no perfect algorithm exists. Your imperfect solution will nonetheless create a fun user experience!

The Road Ahead

We break the notebook into separate steps. Feel free to use the links below to navigate the notebook.

  • Step 0: Import Datasets
  • Step 1: Detect Humans
  • Step 2: Detect Dogs
  • Step 3: Create a CNN to Classify Dog Breeds (from Scratch)
  • Step 4: Use a CNN to Classify Dog Breeds (using Transfer Learning)
  • Step 5: Create a CNN to Classify Dog Breeds (using Transfer Learning)
  • Step 6: Write your Algorithm
  • Step 7: Test Your Algorithm

Step 0: Import Datasets

Import Dog Dataset

In the code cell below, we import a dataset of dog images. We populate a few variables through the use of the load_files function from the scikit-learn library:

  • train_files, valid_files, test_files - numpy arrays containing file paths to images
  • train_targets, valid_targets, test_targets - numpy arrays containing onehot-encoded classification labels
  • dog_names - list of string-valued dog breed names for translating labels
In [1]:
import cv2
import matplotlib.pyplot as plt
%matplotlib inline

def visualize_img(img_path, ax):
    img = cv2.imread(img_path)
    ax.imshow(cv2.cvtColor(img, cv2.COLOR_BGR2RGB))
    
def show_image_grid(files):
    fig = plt.figure(figsize=(20, 10))
    show_images = min(len(files), 12)
    images_per_row = min(len(files), 12)
    for i in range(show_images):
        ax = fig.add_subplot(int(show_images/images_per_row), images_per_row, i + 1, xticks=[], yticks=[])
        visualize_img(files[i], ax)
    plt.show()
In [2]:
from sklearn.datasets import load_files       
from keras.utils import np_utils
import numpy as np
from glob import glob

# define function to load train, test, and validation datasets
def load_dataset(path):
    data = load_files(path)
    dog_files = np.array(data['filenames'])
    dog_targets = np_utils.to_categorical(np.array(data['target']), 133)
    return dog_files, dog_targets

# load train, test, and validation datasets
train_files, train_targets = load_dataset('dogImages/train')
valid_files, valid_targets = load_dataset('dogImages/valid')
test_files, test_targets = load_dataset('dogImages/test')

# load list of dog names
dog_names = [item[20:-1] for item in sorted(glob("dogImages/train/*/"))]

# print statistics about the dataset
print('There are %d total dog categories.' % len(dog_names))
print('There are %s total dog images.\n' % len(np.hstack([train_files, valid_files, test_files])))
print('There are %d training dog images.' % len(train_files))
print('There are %d validation dog images.' % len(valid_files))
print('There are %d test dog images.'% len(test_files))

show_image_grid(train_files)
Using TensorFlow backend.
There are 133 total dog categories.
There are 8351 total dog images.

There are 6680 training dog images.
There are 835 validation dog images.
There are 836 test dog images.

Import Human Dataset

In the code cell below, we import a dataset of human images, where the file paths are stored in the numpy array human_files.

In [3]:
import random
random.seed(8675309)

# load filenames in shuffled human dataset
human_files = np.array(glob("lfw/*/*"))
random.shuffle(human_files)

# print statistics about the dataset
print('There are %d total human images.' % len(human_files))

show_image_grid(human_files)
There are 13233 total human images.

Step 1: Detect Humans

We use OpenCV's implementation of Haar feature-based cascade classifiers to detect human faces in images. OpenCV provides many pre-trained face detectors, stored as XML files on github. We have downloaded one of these detectors and stored it in the haarcascades directory.

In the next code cell, we demonstrate how to use this detector to find human faces in a sample image.

In [4]:
##### import cv2                
import matplotlib.pyplot as plt                        
%matplotlib inline                               

# extract pre-trained face detector
face_cascade = cv2.CascadeClassifier('haarcascades/haarcascade_frontalface_alt.xml')

# load color (BGR) image
img = cv2.imread(human_files[3])
# convert BGR image to grayscale
gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)

# find faces in image
faces = face_cascade.detectMultiScale(gray)

# print number of faces detected in the image
print('Number of faces detected:', len(faces))

# get bounding box for each detected face
for (x,y,w,h) in faces:
    # add bounding box to color image
    cv2.rectangle(img,(x,y),(x+w,y+h),(255,0,0),2)
    
# convert BGR image to RGB for plotting
cv_rgb = cv2.cvtColor(img, cv2.COLOR_BGR2RGB)

# display the image, along with bounding box
plt.imshow(cv_rgb)
plt.show()
Number of faces detected: 1

Before using any of the face detectors, it is standard procedure to convert the images to grayscale. The detectMultiScale function executes the classifier stored in face_cascade and takes the grayscale image as a parameter.

In the above code, faces is a numpy array of detected faces, where each row corresponds to a detected face. Each detected face is a 1D array with four entries that specifies the bounding box of the detected face. The first two entries in the array (extracted in the above code as x and y) specify the horizontal and vertical positions of the top left corner of the bounding box. The last two entries in the array (extracted here as w and h) specify the width and height of the box.

Write a Human Face Detector

We can use this procedure to write a function that returns True if a human face is detected in an image and False otherwise. This function, aptly named face_detector, takes a string-valued file path to an image as input and appears in the code block below.

In [5]:
# returns "True" if face is detected in image stored at img_path

def face_detector(img_path):
    img = cv2.imread(img_path)
    gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
    faces = face_cascade.detectMultiScale(gray)
    return len(faces) > 0

(IMPLEMENTATION) Assess the Human Face Detector

Question 1: Use the code cell below to test the performance of the face_detector function.

  • What percentage of the first 100 images in human_files have a detected human face?
  • What percentage of the first 100 images in dog_files have a detected human face?

Ideally, we would like 100% of human images with a detected face and 0% of dog images with a detected face. You will see that our algorithm falls short of this goal, but still gives acceptable performance. We extract the file paths for the first 100 images from each of the datasets and store them in the numpy arrays human_files_short and dog_files_short.

Answer:

  • 98% of human faces detected correctly
  • 11% of dog faces detected as humans
In [6]:
human_files_short = human_files[:100]
dog_files_short = train_files[:100]
# Do NOT modify the code above this line.

face_detection_sample_size = len(human_files_short)

## TODO: Test the performance of the face_detector algorithm 
## on the images in human_files_short and dog_files_short.
undetected_human_faces = []
undetected_dog_faces = []

for i in range(face_detection_sample_size):
    if face_detector(human_files_short[i]) == 0:
        undetected_human_faces.append(human_files_short[i])
        
    if face_detector(dog_files_short[i]) == 0:
        undetected_dog_faces.append(dog_files_short[i])

print("Percentage of human faces %d%%" % int(face_detection_sample_size - len(undetected_human_faces)))
print("Percentage of dog faces %d%%" % int(face_detection_sample_size - len(undetected_dog_faces)))

if len(undetected_human_faces):
    print("\nMisclassified human faces")
    show_image_grid(undetected_human_faces[:6])
    
if len(undetected_dog_faces):
    print("\nMisclassified dog faces")
    show_image_grid(undetected_dog_faces[:6])
Percentage of human faces 98%
Percentage of dog faces 11%

Misclassified human faces
Misclassified dog faces

Question 2: This algorithmic choice necessitates that we communicate to the user that we accept human images only when they provide a clear view of a face (otherwise, we risk having unneccessarily frustrated users!). In your opinion, is this a reasonable expectation to pose on the user? If not, can you think of a way to detect humans in images that does not necessitate an image with a clearly presented face?

Answer:

  • Haar cascades is an inexpensive option with a simple user requiriment (clear view of a face), which is a resonable expectation, especially for an MVP project. However it is evident that human face detection has problems with obstructed (covered head) and profile images. Dog's images are detected as human faces. Huge entertainment factor. Users can be advised about these limitations, but I think, that Haar cascades approach would not be acceptable as a production version. CNN may offer a better classification efficacy and it is worth exploring in a long run. This fuctionality is well isolated and can be improved later without a risk of affecting other components.

We suggest the face detector from OpenCV as a potential way to detect human images in your algorithm, but you are free to explore other approaches, especially approaches that make use of deep learning :). Please use the code cell below to design and test your own face detection algorithm. If you decide to pursue this optional task, report performance on each of the datasets.

(Optional) TODO: Report the performance of another

face detection algorithm on the LFW dataset

Feel free to use as many code cells as needed.


Step 2: Detect Dogs

In this section, we use a pre-trained ResNet-50 model to detect dogs in images. Our first line of code downloads the ResNet-50 model, along with weights that have been trained on ImageNet, a very large, very popular dataset used for image classification and other vision tasks. ImageNet contains over 10 million URLs, each linking to an image containing an object from one of 1000 categories. Given an image, this pre-trained ResNet-50 model returns a prediction (derived from the available categories in ImageNet) for the object that is contained in the image.

In [7]:
from keras.applications.resnet50 import ResNet50

# define ResNet50 model
ResNet50_model = ResNet50(weights='imagenet')

Pre-process the Data

When using TensorFlow as backend, Keras CNNs require a 4D array (which we'll also refer to as a 4D tensor) as input, with shape

$$ (\text{nb_samples}, \text{rows}, \text{columns}, \text{channels}), $$

where nb_samples corresponds to the total number of images (or samples), and rows, columns, and channels correspond to the number of rows, columns, and channels for each image, respectively.

The path_to_tensor function below takes a string-valued file path to a color image as input and returns a 4D tensor suitable for supplying to a Keras CNN. The function first loads the image and resizes it to a square image that is $224 \times 224$ pixels. Next, the image is converted to an array, which is then resized to a 4D tensor. In this case, since we are working with color images, each image has three channels. Likewise, since we are processing a single image (or sample), the returned tensor will always have shape

$$ (1, 224, 224, 3). $$

The paths_to_tensor function takes a numpy array of string-valued image paths as input and returns a 4D tensor with shape

$$ (\text{nb_samples}, 224, 224, 3). $$

Here, nb_samples is the number of samples, or number of images, in the supplied array of image paths. It is best to think of nb_samples as the number of 3D tensors (where each 3D tensor corresponds to a different image) in your dataset!

In [8]:
from keras.preprocessing import image                  
from tqdm import tqdm

def path_to_tensor(img_path):
    # loads RGB image as PIL.Image.Image type
    img = image.load_img(img_path, target_size=(224, 224))
    # convert PIL.Image.Image type to 3D tensor with shape (224, 224, 3)
    x = image.img_to_array(img)
    # convert 3D tensor to 4D tensor with shape (1, 224, 224, 3) and return 4D tensor
    return np.expand_dims(x, axis=0)

def paths_to_tensor(img_paths):
    list_of_tensors = [path_to_tensor(img_path) for img_path in tqdm(img_paths)]
    return np.vstack(list_of_tensors)

Making Predictions with ResNet-50

Getting the 4D tensor ready for ResNet-50, and for any other pre-trained model in Keras, requires some additional processing. First, the RGB image is converted to BGR by reordering the channels. All pre-trained models have the additional normalization step that the mean pixel (expressed in RGB as $[103.939, 116.779, 123.68]$ and calculated from all pixels in all images in ImageNet) must be subtracted from every pixel in each image. This is implemented in the imported function preprocess_input. If you're curious, you can check the code for preprocess_input here.

Now that we have a way to format our image for supplying to ResNet-50, we are now ready to use the model to extract the predictions. This is accomplished with the predict method, which returns an array whose $i$-th entry is the model's predicted probability that the image belongs to the $i$-th ImageNet category. This is implemented in the ResNet50_predict_labels function below.

By taking the argmax of the predicted probability vector, we obtain an integer corresponding to the model's predicted object class, which we can identify with an object category through the use of this dictionary.

In [9]:
from keras.applications.resnet50 import preprocess_input, decode_predictions

def ResNet50_predict_labels(img_path):
    # returns prediction vector for image located at img_path
    img = preprocess_input(path_to_tensor(img_path))
    return np.argmax(ResNet50_model.predict(img))

Write a Dog Detector

While looking at the dictionary, you will notice that the categories corresponding to dogs appear in an uninterrupted sequence and correspond to dictionary keys 151-268, inclusive, to include all categories from 'Chihuahua' to 'Mexican hairless'. Thus, in order to check to see if an image is predicted to contain a dog by the pre-trained ResNet-50 model, we need only check if the ResNet50_predict_labels function above returns a value between 151 and 268 (inclusive).

We use these ideas to complete the dog_detector function below, which returns True if a dog is detected in an image (and False if not).

In [10]:
### returns "True" if a dog is detected in the image stored at img_path
def dog_detector(img_path):
    prediction = ResNet50_predict_labels(img_path)
    return ((prediction <= 268) & (prediction >= 151)) 

(IMPLEMENTATION) Assess the Dog Detector

Question 3: Use the code cell below to test the performance of your dog_detector function.

  • What percentage of the images in human_files_short have a detected dog?
  • What percentage of the images in dog_files_short have a detected dog?

Answer:

  • 2% of human faces detected as dogs
  • 100% of dog faces detected correctly
In [11]:
### TODO: Test the performance of the dog_detector function
### on the images in human_files_short and dog_files_short.

undetected_human_faces = []
undetected_dog_faces = []

for i in range(face_detection_sample_size):
    if dog_detector(human_files_short[i]) == 0:
        undetected_human_faces.append(human_files_short[i])
        
    if dog_detector(dog_files_short[i]) == 0:
        undetected_dog_faces.append(dog_files_short[i])

print("Percentage of human faces %d%%" % int(face_detection_sample_size - len(undetected_human_faces)))
print("Percentage of dog faces %d%%" % int(face_detection_sample_size - len(undetected_dog_faces)))

if len(undetected_human_faces):
    print("\nMisclassified human faces")
    show_image_grid(undetected_human_faces[:6])
    
if len(undetected_dog_faces):
    print("\nMisclassified dog faces")
    show_image_grid(undetected_dog_faces[:6])
Percentage of human faces 2%
Percentage of dog faces 100%

Misclassified human faces

Step 3: Create a CNN to Classify Dog Breeds (from Scratch)

Now that we have functions for detecting humans and dogs in images, we need a way to predict breed from images. In this step, you will create a CNN that classifies dog breeds. You must create your CNN from scratch (so, you can't use transfer learning yet!), and you must attain a test accuracy of at least 1%. In Step 5 of this notebook, you will have the opportunity to use transfer learning to create a CNN that attains greatly improved accuracy.

Be careful with adding too many trainable layers! More parameters means longer training, which means you are more likely to need a GPU to accelerate the training process. Thankfully, Keras provides a handy estimate of the time that each epoch is likely to take; you can extrapolate this estimate to figure out how long it will take for your algorithm to train.

We mention that the task of assigning breed to dogs from images is considered exceptionally challenging. To see why, consider that even a human would have great difficulty in distinguishing between a Brittany and a Welsh Springer Spaniel.

Brittany Welsh Springer Spaniel

It is not difficult to find other dog breed pairs with minimal inter-class variation (for instance, Curly-Coated Retrievers and American Water Spaniels).

Curly-Coated Retriever American Water Spaniel

Likewise, recall that labradors come in yellow, chocolate, and black. Your vision-based algorithm will have to conquer this high intra-class variation to determine how to classify all of these different shades as the same breed.

Yellow Labrador Chocolate Labrador Black Labrador

We also mention that random chance presents an exceptionally low bar: setting aside the fact that the classes are slightly imabalanced, a random guess will provide a correct answer roughly 1 in 133 times, which corresponds to an accuracy of less than 1%.

Remember that the practice is far ahead of the theory in deep learning. Experiment with many different architectures, and trust your intuition. And, of course, have fun!

Pre-process the Data

We rescale the images by dividing every pixel in every image by 255.

In [12]:
from PIL import ImageFile                            
ImageFile.LOAD_TRUNCATED_IMAGES = True                 

# pre-process the data for Keras
train_tensors = paths_to_tensor(train_files).astype('float32')/255
valid_tensors = paths_to_tensor(valid_files).astype('float32')/255
test_tensors = paths_to_tensor(test_files).astype('float32')/255
100%|██████████| 6680/6680 [00:53<00:00, 124.51it/s]
100%|██████████| 835/835 [00:06<00:00, 138.52it/s]
100%|██████████| 836/836 [00:06<00:00, 139.27it/s]

(IMPLEMENTATION) Model Architecture

Create a CNN to classify dog breed. At the end of your code cell block, summarize the layers of your model by executing the line:

    model.summary()

We have imported some Python modules to get you started, but feel free to import as many modules as you need. If you end up getting stuck, here's a hint that specifies a model that trains relatively fast on CPU and attains >1% test accuracy in 5 epochs:

Sample CNN

Question 4: Outline the steps you took to get to your final CNN architecture and your reasoning at each step. If you chose to use the hinted architecture above, describe why you think that CNN architecture should work well for the image classification task.

Answer:

I have chosen hinted architecture since it is know to work well for image classification by using several convolutional layers with exponentially increasing number of filters (16, 32, 64, 128, 256, 512). Early filters learn basic graphical features. Later ones learn more advanced ones specific to our training set. Max pooling provide dimention reduction to keep trainiing time reasonable. Batch normalization is used to increase independnce of each hidden layer and improve model generalization. Dropout step has been added to avoid overfiting. Output layer is dense layer of 133 nodes to predicts final classes. The model is a result of series of experiments that have lead to the current architecture.

In [13]:
from keras.layers import Conv2D, MaxPooling2D, GlobalAveragePooling2D
from keras.layers import Dropout, Flatten, Dense
from keras.models import Sequential
from keras.layers.normalization import BatchNormalization

model = Sequential()

### TODO: Define your architecture.

model.add(Conv2D(filters=16, kernel_size=2, activation='relu', input_shape=(224, 224, 3)))
model.add(MaxPooling2D(pool_size=2))
model.add(BatchNormalization())

model.add(Conv2D(filters=32, kernel_size=2, activation='relu', kernel_initializer='he_normal'))
model.add(MaxPooling2D(pool_size=2))
model.add(BatchNormalization())

model.add(Conv2D(filters=64, kernel_size=2, activation='relu', kernel_initializer='he_normal'))
model.add(MaxPooling2D(pool_size=2))
model.add(BatchNormalization())

model.add(Conv2D(filters=128, kernel_size=2, activation='relu', kernel_initializer='he_normal'))
model.add(MaxPooling2D(pool_size=2))
model.add(BatchNormalization())

model.add(Conv2D(filters=256, kernel_size=2, activation='relu', kernel_initializer='he_normal'))
model.add(MaxPooling2D(pool_size=2))
model.add(BatchNormalization())

model.add(Conv2D(filters=512, kernel_size=2, activation='relu', kernel_initializer='he_normal'))
model.add(MaxPooling2D(pool_size=2))
model.add(BatchNormalization())

model.add(Dropout(0.3))

model.add(GlobalAveragePooling2D())
model.add(Dense(133, activation='softmax'))

model.summary()
_________________________________________________________________
Layer (type)                 Output Shape              Param #   
=================================================================
conv2d_1 (Conv2D)            (None, 223, 223, 16)      208       
_________________________________________________________________
max_pooling2d_2 (MaxPooling2 (None, 111, 111, 16)      0         
_________________________________________________________________
batch_normalization_1 (Batch (None, 111, 111, 16)      64        
_________________________________________________________________
conv2d_2 (Conv2D)            (None, 110, 110, 32)      2080      
_________________________________________________________________
max_pooling2d_3 (MaxPooling2 (None, 55, 55, 32)        0         
_________________________________________________________________
batch_normalization_2 (Batch (None, 55, 55, 32)        128       
_________________________________________________________________
conv2d_3 (Conv2D)            (None, 54, 54, 64)        8256      
_________________________________________________________________
max_pooling2d_4 (MaxPooling2 (None, 27, 27, 64)        0         
_________________________________________________________________
batch_normalization_3 (Batch (None, 27, 27, 64)        256       
_________________________________________________________________
conv2d_4 (Conv2D)            (None, 26, 26, 128)       32896     
_________________________________________________________________
max_pooling2d_5 (MaxPooling2 (None, 13, 13, 128)       0         
_________________________________________________________________
batch_normalization_4 (Batch (None, 13, 13, 128)       512       
_________________________________________________________________
conv2d_5 (Conv2D)            (None, 12, 12, 256)       131328    
_________________________________________________________________
max_pooling2d_6 (MaxPooling2 (None, 6, 6, 256)         0         
_________________________________________________________________
batch_normalization_5 (Batch (None, 6, 6, 256)         1024      
_________________________________________________________________
conv2d_6 (Conv2D)            (None, 5, 5, 512)         524800    
_________________________________________________________________
max_pooling2d_7 (MaxPooling2 (None, 2, 2, 512)         0         
_________________________________________________________________
batch_normalization_6 (Batch (None, 2, 2, 512)         2048      
_________________________________________________________________
dropout_1 (Dropout)          (None, 2, 2, 512)         0         
_________________________________________________________________
global_average_pooling2d_1 ( (None, 512)               0         
_________________________________________________________________
dense_1 (Dense)              (None, 133)               68229     
=================================================================
Total params: 771,829.0
Trainable params: 769,813.0
Non-trainable params: 2,016.0
_________________________________________________________________

Compile the Model

In [14]:
model.compile(optimizer='rmsprop', loss='categorical_crossentropy', metrics=['accuracy'])

(IMPLEMENTATION) Train the Model

Train your model in the code cell below. Use model checkpointing to save the model that attains the best validation loss.

You are welcome to augment the training data, but this is not a requirement.

In [15]:
from keras.callbacks import ModelCheckpoint, EarlyStopping  

### TODO: specify the number of epochs that you would like to use to train the model.

from keras.preprocessing.image import ImageDataGenerator

# create and configure augmented image generator
datagen = ImageDataGenerator(
    rotation_range=10,
    zoom_range=0.2,
    width_shift_range=0.1,
    height_shift_range=0.1,
    horizontal_flip=True)

# fit augmented image generator on data
datagen.fit(train_tensors)

epochs = 100

### Do NOT modify the code below this line.

checkpointer = ModelCheckpoint(filepath='saved_models/weights.best.from_scratch.hdf5', 
                               verbose=1, save_best_only=True)

# Stop the training if the model shows no improvement 
stopper = EarlyStopping(monitor='val_loss', min_delta=0.05, patience=3, verbose=1, mode='auto')

model.fit(train_tensors, train_targets, 
          validation_data=(valid_tensors, valid_targets),
          epochs=epochs, batch_size=20, 
          callbacks=[checkpointer, stopper], verbose=1)
Train on 6680 samples, validate on 835 samples
Epoch 1/100
6660/6680 [============================>.] - ETA: 0s - loss: 4.5767 - acc: 0.0556Epoch 00000: val_loss improved from inf to 5.15430, saving model to saved_models/weights.best.from_scratch.hdf5
6680/6680 [==============================] - 30s - loss: 4.5740 - acc: 0.0557 - val_loss: 5.1543 - val_acc: 0.0144
Epoch 2/100
6660/6680 [============================>.] - ETA: 0s - loss: 3.9062 - acc: 0.1207Epoch 00001: val_loss improved from 5.15430 to 4.02507, saving model to saved_models/weights.best.from_scratch.hdf5
6680/6680 [==============================] - 28s - loss: 3.9060 - acc: 0.1207 - val_loss: 4.0251 - val_acc: 0.1030
Epoch 3/100
6660/6680 [============================>.] - ETA: 0s - loss: 3.4863 - acc: 0.1964Epoch 00002: val_loss improved from 4.02507 to 3.70878, saving model to saved_models/weights.best.from_scratch.hdf5
6680/6680 [==============================] - 28s - loss: 3.4856 - acc: 0.1964 - val_loss: 3.7088 - val_acc: 0.1425
Epoch 4/100
6660/6680 [============================>.] - ETA: 0s - loss: 3.0595 - acc: 0.2725Epoch 00003: val_loss improved from 3.70878 to 3.49870, saving model to saved_models/weights.best.from_scratch.hdf5
6680/6680 [==============================] - 28s - loss: 3.0599 - acc: 0.2720 - val_loss: 3.4987 - val_acc: 0.1784
Epoch 5/100
6660/6680 [============================>.] - ETA: 0s - loss: 2.6122 - acc: 0.3856Epoch 00004: val_loss did not improve
6680/6680 [==============================] - 28s - loss: 2.6129 - acc: 0.3853 - val_loss: 3.6791 - val_acc: 0.1593
Epoch 6/100
6660/6680 [============================>.] - ETA: 0s - loss: 2.1174 - acc: 0.5117Epoch 00005: val_loss did not improve
6680/6680 [==============================] - 28s - loss: 2.1168 - acc: 0.5117 - val_loss: 3.6593 - val_acc: 0.1844
Epoch 7/100
6660/6680 [============================>.] - ETA: 0s - loss: 1.6432 - acc: 0.6476Epoch 00006: val_loss improved from 3.49870 to 3.22338, saving model to saved_models/weights.best.from_scratch.hdf5
6680/6680 [==============================] - 28s - loss: 1.6420 - acc: 0.6481 - val_loss: 3.2234 - val_acc: 0.2395
Epoch 8/100
6660/6680 [============================>.] - ETA: 0s - loss: 1.2041 - acc: 0.7575Epoch 00007: val_loss did not improve
6680/6680 [==============================] - 28s - loss: 1.2046 - acc: 0.7572 - val_loss: 3.3544 - val_acc: 0.2443
Epoch 9/100
6660/6680 [============================>.] - ETA: 0s - loss: 0.8148 - acc: 0.8511Epoch 00008: val_loss did not improve
6680/6680 [==============================] - 28s - loss: 0.8160 - acc: 0.8507 - val_loss: 3.3612 - val_acc: 0.2311
Epoch 10/100
6660/6680 [============================>.] - ETA: 0s - loss: 0.5382 - acc: 0.9116Epoch 00009: val_loss did not improve
6680/6680 [==============================] - 28s - loss: 0.5375 - acc: 0.9117 - val_loss: 3.5247 - val_acc: 0.2275
Epoch 11/100
6660/6680 [============================>.] - ETA: 0s - loss: 0.3777 - acc: 0.9432Epoch 00010: val_loss did not improve
6680/6680 [==============================] - 28s - loss: 0.3775 - acc: 0.9433 - val_loss: 3.3950 - val_acc: 0.2671
Epoch 00010: early stopping
Out[15]:
<keras.callbacks.History at 0x7f3d804d5d30>

Load the Model with the Best Validation Loss

In [16]:
model.load_weights('saved_models/weights.best.from_scratch.hdf5')

Test the Model

Try out your model on the test dataset of dog images. Ensure that your test accuracy is greater than 1%.

In [17]:
# get index of predicted dog breed for each image in test set
dog_breed_predictions = [np.argmax(model.predict(np.expand_dims(tensor, axis=0))) for tensor in test_tensors]

# report test accuracy
test_accuracy = 100*np.sum(np.array(dog_breed_predictions)==np.argmax(test_targets, axis=1))/len(dog_breed_predictions)
print('Test accuracy: %.4f%%' % test_accuracy)
Test accuracy: 25.1196%

Step 4: Use a CNN to Classify Dog Breeds

To reduce training time without sacrificing accuracy, we show you how to train a CNN using transfer learning. In the following step, you will get a chance to use transfer learning to train your own CNN.

Obtain Bottleneck Features

In [18]:
bottleneck_features = np.load('bottleneck_features/DogVGG16Data.npz')
train_VGG16 = bottleneck_features['train']
valid_VGG16 = bottleneck_features['valid']
test_VGG16 = bottleneck_features['test']

Model Architecture

The model uses the the pre-trained VGG-16 model as a fixed feature extractor, where the last convolutional output of VGG-16 is fed as input to our model. We only add a global average pooling layer and a fully connected layer, where the latter contains one node for each dog category and is equipped with a softmax.

In [19]:
VGG16_model = Sequential()
VGG16_model.add(GlobalAveragePooling2D(input_shape=train_VGG16.shape[1:]))
VGG16_model.add(Dense(133, activation='softmax'))

VGG16_model.summary()
_________________________________________________________________
Layer (type)                 Output Shape              Param #   
=================================================================
global_average_pooling2d_2 ( (None, 512)               0         
_________________________________________________________________
dense_2 (Dense)              (None, 133)               68229     
=================================================================
Total params: 68,229.0
Trainable params: 68,229.0
Non-trainable params: 0.0
_________________________________________________________________

Compile the Model

In [20]:
VGG16_model.compile(loss='categorical_crossentropy', optimizer='rmsprop', metrics=['accuracy'])

Train the Model

In [21]:
checkpointer = ModelCheckpoint(filepath='saved_models/weights.best.VGG16.hdf5', 
                               verbose=1, save_best_only=True)

VGG16_model.fit(train_VGG16, train_targets, 
          validation_data=(valid_VGG16, valid_targets),
          epochs=20, batch_size=20, callbacks=[checkpointer], verbose=1)
Train on 6680 samples, validate on 835 samples
Epoch 1/20
6500/6680 [============================>.] - ETA: 0s - loss: 12.0978 - acc: 0.1382Epoch 00000: val_loss improved from inf to 10.84167, saving model to saved_models/weights.best.VGG16.hdf5
6680/6680 [==============================] - 1s - loss: 12.0443 - acc: 0.1409 - val_loss: 10.8417 - val_acc: 0.2120
Epoch 2/20
6500/6680 [============================>.] - ETA: 0s - loss: 10.4017 - acc: 0.2740Epoch 00001: val_loss improved from 10.84167 to 10.46394, saving model to saved_models/weights.best.VGG16.hdf5
6680/6680 [==============================] - 1s - loss: 10.3981 - acc: 0.2740 - val_loss: 10.4639 - val_acc: 0.2814
Epoch 3/20
6420/6680 [===========================>..] - ETA: 0s - loss: 10.0596 - acc: 0.3262Epoch 00002: val_loss improved from 10.46394 to 10.31218, saving model to saved_models/weights.best.VGG16.hdf5
6680/6680 [==============================] - 1s - loss: 10.0498 - acc: 0.3275 - val_loss: 10.3122 - val_acc: 0.2994
Epoch 4/20
6480/6680 [============================>.] - ETA: 0s - loss: 9.8807 - acc: 0.3545Epoch 00003: val_loss improved from 10.31218 to 10.28407, saving model to saved_models/weights.best.VGG16.hdf5
6680/6680 [==============================] - 1s - loss: 9.8920 - acc: 0.3540 - val_loss: 10.2841 - val_acc: 0.3066
Epoch 5/20
6620/6680 [============================>.] - ETA: 0s - loss: 9.7364 - acc: 0.3654Epoch 00004: val_loss improved from 10.28407 to 10.01579, saving model to saved_models/weights.best.VGG16.hdf5
6680/6680 [==============================] - 1s - loss: 9.7271 - acc: 0.3657 - val_loss: 10.0158 - val_acc: 0.3222
Epoch 6/20
6460/6680 [============================>.] - ETA: 0s - loss: 9.4239 - acc: 0.3890Epoch 00005: val_loss improved from 10.01579 to 9.79415, saving model to saved_models/weights.best.VGG16.hdf5
6680/6680 [==============================] - 1s - loss: 9.4350 - acc: 0.3883 - val_loss: 9.7941 - val_acc: 0.3353
Epoch 7/20
6440/6680 [===========================>..] - ETA: 0s - loss: 9.2672 - acc: 0.4016Epoch 00006: val_loss improved from 9.79415 to 9.68836, saving model to saved_models/weights.best.VGG16.hdf5
6680/6680 [==============================] - 1s - loss: 9.2516 - acc: 0.4025 - val_loss: 9.6884 - val_acc: 0.3437
Epoch 8/20
6580/6680 [============================>.] - ETA: 0s - loss: 9.0965 - acc: 0.4146Epoch 00007: val_loss improved from 9.68836 to 9.64495, saving model to saved_models/weights.best.VGG16.hdf5
6680/6680 [==============================] - 1s - loss: 9.0918 - acc: 0.4148 - val_loss: 9.6449 - val_acc: 0.3377
Epoch 9/20
6600/6680 [============================>.] - ETA: 0s - loss: 8.9857 - acc: 0.4264Epoch 00008: val_loss improved from 9.64495 to 9.55031, saving model to saved_models/weights.best.VGG16.hdf5
6680/6680 [==============================] - 1s - loss: 8.9762 - acc: 0.4265 - val_loss: 9.5503 - val_acc: 0.3473
Epoch 10/20
6600/6680 [============================>.] - ETA: 0s - loss: 8.9036 - acc: 0.4355Epoch 00009: val_loss improved from 9.55031 to 9.51726, saving model to saved_models/weights.best.VGG16.hdf5
6680/6680 [==============================] - 1s - loss: 8.9041 - acc: 0.4352 - val_loss: 9.5173 - val_acc: 0.3509
Epoch 11/20
6600/6680 [============================>.] - ETA: 0s - loss: 8.7046 - acc: 0.4405Epoch 00010: val_loss improved from 9.51726 to 9.31605, saving model to saved_models/weights.best.VGG16.hdf5
6680/6680 [==============================] - 1s - loss: 8.7007 - acc: 0.4407 - val_loss: 9.3161 - val_acc: 0.3641
Epoch 12/20
6560/6680 [============================>.] - ETA: 0s - loss: 8.5646 - acc: 0.4488Epoch 00011: val_loss improved from 9.31605 to 9.17031, saving model to saved_models/weights.best.VGG16.hdf5
6680/6680 [==============================] - 1s - loss: 8.5643 - acc: 0.4487 - val_loss: 9.1703 - val_acc: 0.3737
Epoch 13/20
6440/6680 [===========================>..] - ETA: 0s - loss: 8.4063 - acc: 0.4637Epoch 00012: val_loss improved from 9.17031 to 8.99988, saving model to saved_models/weights.best.VGG16.hdf5
6680/6680 [==============================] - 1s - loss: 8.3852 - acc: 0.4647 - val_loss: 8.9999 - val_acc: 0.3772
Epoch 14/20
6620/6680 [============================>.] - ETA: 0s - loss: 8.2005 - acc: 0.4770Epoch 00013: val_loss improved from 8.99988 to 8.90085, saving model to saved_models/weights.best.VGG16.hdf5
6680/6680 [==============================] - 1s - loss: 8.2185 - acc: 0.4760 - val_loss: 8.9008 - val_acc: 0.3940
Epoch 15/20
6660/6680 [============================>.] - ETA: 0s - loss: 8.1571 - acc: 0.4836Epoch 00014: val_loss did not improve
6680/6680 [==============================] - 1s - loss: 8.1689 - acc: 0.4829 - val_loss: 9.0769 - val_acc: 0.3832
Epoch 16/20
6440/6680 [===========================>..] - ETA: 0s - loss: 8.1154 - acc: 0.4843Epoch 00015: val_loss improved from 8.90085 to 8.78414, saving model to saved_models/weights.best.VGG16.hdf5
6680/6680 [==============================] - 1s - loss: 8.0750 - acc: 0.4865 - val_loss: 8.7841 - val_acc: 0.3844
Epoch 17/20
6580/6680 [============================>.] - ETA: 0s - loss: 7.9264 - acc: 0.4938Epoch 00016: val_loss improved from 8.78414 to 8.70671, saving model to saved_models/weights.best.VGG16.hdf5
6680/6680 [==============================] - 1s - loss: 7.9143 - acc: 0.4946 - val_loss: 8.7067 - val_acc: 0.4060
Epoch 18/20
6420/6680 [===========================>..] - ETA: 0s - loss: 7.8247 - acc: 0.5050Epoch 00017: val_loss improved from 8.70671 to 8.58472, saving model to saved_models/weights.best.VGG16.hdf5
6680/6680 [==============================] - 1s - loss: 7.8197 - acc: 0.5051 - val_loss: 8.5847 - val_acc: 0.4060
Epoch 19/20
6560/6680 [============================>.] - ETA: 0s - loss: 7.7259 - acc: 0.5107Epoch 00018: val_loss improved from 8.58472 to 8.46000, saving model to saved_models/weights.best.VGG16.hdf5
6680/6680 [==============================] - 1s - loss: 7.7120 - acc: 0.5111 - val_loss: 8.4600 - val_acc: 0.4084
Epoch 20/20
6600/6680 [============================>.] - ETA: 0s - loss: 7.6244 - acc: 0.5177Epoch 00019: val_loss improved from 8.46000 to 8.28667, saving model to saved_models/weights.best.VGG16.hdf5
6680/6680 [==============================] - 1s - loss: 7.6128 - acc: 0.5184 - val_loss: 8.2867 - val_acc: 0.4180
Out[21]:
<keras.callbacks.History at 0x7f3d802dfd30>

Load the Model with the Best Validation Loss

In [22]:
VGG16_model.load_weights('saved_models/weights.best.VGG16.hdf5')

Test the Model

Now, we can use the CNN to test how well it identifies breed within our test dataset of dog images. We print the test accuracy below.

In [23]:
# get index of predicted dog breed for each image in test set
VGG16_predictions = [np.argmax(VGG16_model.predict(np.expand_dims(feature, axis=0))) for feature in test_VGG16]

# report test accuracy
test_accuracy = 100*np.sum(np.array(VGG16_predictions)==np.argmax(test_targets, axis=1))/len(VGG16_predictions)
print('Test accuracy: %.4f%%' % test_accuracy)
Test accuracy: 42.7033%

Predict Dog Breed with the Model

In [24]:
from extract_bottleneck_features import *

def VGG16_predict_breed(img_path):
    # extract bottleneck features
    bottleneck_feature = extract_VGG16(path_to_tensor(img_path))
    # obtain predicted vector
    predicted_vector = VGG16_model.predict(bottleneck_feature)
    # return dog breed that is predicted by the model
    return dog_names[np.argmax(predicted_vector)]

Step 5: Create a CNN to Classify Dog Breeds (using Transfer Learning)

You will now use transfer learning to create a CNN that can identify dog breed from images. Your CNN must attain at least 60% accuracy on the test set.

In Step 4, we used transfer learning to create a CNN using VGG-16 bottleneck features. In this section, you must use the bottleneck features from a different pre-trained model. To make things easier for you, we have pre-computed the features for all of the networks that are currently available in Keras:

The files are encoded as such:

Dog{network}Data.npz

where {network}, in the above filename, can be one of VGG19, Resnet50, InceptionV3, or Xception. Pick one of the above architectures, download the corresponding bottleneck features, and store the downloaded file in the bottleneck_features/ folder in the repository.

(IMPLEMENTATION) Obtain Bottleneck Features

In the code block below, extract the bottleneck features corresponding to the train, test, and validation sets by running the following:

bottleneck_features = np.load('bottleneck_features/Dog{network}Data.npz')
train_{network} = bottleneck_features['train']
valid_{network} = bottleneck_features['valid']
test_{network} = bottleneck_features['test']
In [25]:
##### TODO: Obtain bottleneck features from another pre-trained CNN.
bottleneck_features = np.load('bottleneck_features/DogXceptionData.npz')
train_Xception = bottleneck_features['train']
valid_Xception = bottleneck_features['valid']
test_Xception = bottleneck_features['test']

#from keras.applications.vgg19 import VGG19; VGG19().summary()
bottleneck_features = np.load('bottleneck_features/DogVGG19Data.npz')
train_VGG19 = bottleneck_features['train']
valid_VGG19 = bottleneck_features['valid']
test_VGG19 = bottleneck_features['test']

bottleneck_features = np.load('bottleneck_features/DogResnet50Data.npz')
train_Resnet50 = bottleneck_features['train']
valid_Resnet50 = bottleneck_features['valid']
test_Resnet50 = bottleneck_features['test']
In [26]:
train_transfer = train_Xception
valid_transfer = valid_Xception
test_transfer = test_Xception

(IMPLEMENTATION) Model Architecture

Create a CNN to classify dog breed. At the end of your code cell block, summarize the layers of your model by executing the line:

    <your model's name>.summary()

Question 5: Outline the steps you took to get to your final CNN architecture and your reasoning at each step. Describe why you think the architecture is suitable for the current problem.

Answer:

Transfer learning technique is used for proposed classificator allowing us to define very simple architechure. The architechure consists of two layers. GlobalAveragePooling2D layer is used to avoid overfiting and reduce data dimentionality. Experiments showed significanly better perfomance of GlobalAveragePooling2D vs Flatten for givven bootleneck features and training data. Final Dense fully connected layer with softmax activation is used to output probabilities of breeds we are see to classify

In [27]:
### TODO: Define your architecture.
tmodel = Sequential()
tmodel.add(GlobalAveragePooling2D(input_shape=train_transfer.shape[1:]))
tmodel.add(Dense(133, activation='softmax'))
tmodel.summary()
_________________________________________________________________
Layer (type)                 Output Shape              Param #   
=================================================================
global_average_pooling2d_3 ( (None, 2048)              0         
_________________________________________________________________
dense_3 (Dense)              (None, 133)               272517    
=================================================================
Total params: 272,517.0
Trainable params: 272,517.0
Non-trainable params: 0.0
_________________________________________________________________

(IMPLEMENTATION) Compile the Model

In [28]:
### TODO: Compile the model.
tmodel.compile(loss='categorical_crossentropy', optimizer='rmsprop', 
                  metrics=['accuracy'])

(IMPLEMENTATION) Train the Model

Train your model in the code cell below. Use model checkpointing to save the model that attains the best validation loss.

You are welcome to augment the training data, but this is not a requirement.

In [29]:
### TODO: Train the model.
from keras.callbacks import ModelCheckpoint, EarlyStopping  

epochs = 100

# Stop the training if the model shows no improvement 
stopper = EarlyStopping(monitor='val_loss', min_delta=0.01, patience=10, verbose=1, mode='auto')

### Do NOT modify the code below this line.

checkpointer = ModelCheckpoint(filepath='saved_models/weights.transfer.best.hdf5', 
                               verbose=1, save_best_only=True)

tmodel.fit(train_transfer, train_targets, 
          validation_data=(valid_transfer, valid_targets),
          epochs=epochs, batch_size=20, 
          callbacks=[checkpointer, stopper], verbose=1)
Train on 6680 samples, validate on 835 samples
Epoch 1/100
6640/6680 [============================>.] - ETA: 0s - loss: 1.0583 - acc: 0.7378Epoch 00000: val_loss improved from inf to 0.52456, saving model to saved_models/weights.transfer.best.hdf5
6680/6680 [==============================] - 3s - loss: 1.0542 - acc: 0.7388 - val_loss: 0.5246 - val_acc: 0.8228
Epoch 2/100
6660/6680 [============================>.] - ETA: 0s - loss: 0.3961 - acc: 0.8752Epoch 00001: val_loss improved from 0.52456 to 0.49223, saving model to saved_models/weights.transfer.best.hdf5
6680/6680 [==============================] - 3s - loss: 0.3967 - acc: 0.8750 - val_loss: 0.4922 - val_acc: 0.8323
Epoch 3/100
6620/6680 [============================>.] - ETA: 0s - loss: 0.3153 - acc: 0.8983Epoch 00002: val_loss did not improve
6680/6680 [==============================] - 3s - loss: 0.3151 - acc: 0.8984 - val_loss: 0.5062 - val_acc: 0.8479
Epoch 4/100
6580/6680 [============================>.] - ETA: 0s - loss: 0.2795 - acc: 0.9135Epoch 00003: val_loss did not improve
6680/6680 [==============================] - 3s - loss: 0.2796 - acc: 0.9136 - val_loss: 0.5580 - val_acc: 0.8395
Epoch 5/100
6640/6680 [============================>.] - ETA: 0s - loss: 0.2448 - acc: 0.9241Epoch 00004: val_loss did not improve
6680/6680 [==============================] - 3s - loss: 0.2447 - acc: 0.9241 - val_loss: 0.5215 - val_acc: 0.8491
Epoch 6/100
6560/6680 [============================>.] - ETA: 0s - loss: 0.2152 - acc: 0.9338Epoch 00005: val_loss did not improve
6680/6680 [==============================] - 3s - loss: 0.2186 - acc: 0.9328 - val_loss: 0.5252 - val_acc: 0.8515
Epoch 7/100
6640/6680 [============================>.] - ETA: 0s - loss: 0.1948 - acc: 0.9389Epoch 00006: val_loss did not improve
6680/6680 [==============================] - 3s - loss: 0.1951 - acc: 0.9388 - val_loss: 0.5316 - val_acc: 0.8539
Epoch 8/100
6660/6680 [============================>.] - ETA: 0s - loss: 0.1783 - acc: 0.9458Epoch 00007: val_loss did not improve
6680/6680 [==============================] - 3s - loss: 0.1778 - acc: 0.9460 - val_loss: 0.5561 - val_acc: 0.8575
Epoch 9/100
6660/6680 [============================>.] - ETA: 0s - loss: 0.1610 - acc: 0.9515Epoch 00008: val_loss did not improve
6680/6680 [==============================] - 3s - loss: 0.1612 - acc: 0.9515 - val_loss: 0.5614 - val_acc: 0.8539
Epoch 10/100
6640/6680 [============================>.] - ETA: 0s - loss: 0.1473 - acc: 0.9553Epoch 00009: val_loss did not improve
6680/6680 [==============================] - 3s - loss: 0.1479 - acc: 0.9551 - val_loss: 0.5841 - val_acc: 0.8539
Epoch 11/100
6560/6680 [============================>.] - ETA: 0s - loss: 0.1339 - acc: 0.9599Epoch 00010: val_loss did not improve
6680/6680 [==============================] - 3s - loss: 0.1330 - acc: 0.9602 - val_loss: 0.5687 - val_acc: 0.8551
Epoch 12/100
6660/6680 [============================>.] - ETA: 0s - loss: 0.1254 - acc: 0.9611Epoch 00011: val_loss did not improve
6680/6680 [==============================] - 3s - loss: 0.1252 - acc: 0.9612 - val_loss: 0.6009 - val_acc: 0.8563
Epoch 13/100
6640/6680 [============================>.] - ETA: 0s - loss: 0.1143 - acc: 0.9670Epoch 00012: val_loss did not improve
6680/6680 [==============================] - 3s - loss: 0.1139 - acc: 0.9672 - val_loss: 0.6296 - val_acc: 0.8455
Epoch 00012: early stopping
Out[29]:
<keras.callbacks.History at 0x7f3d8014d908>

(IMPLEMENTATION) Load the Model with the Best Validation Loss

In [30]:
### TODO: Load the model weights with the best validation loss.
tmodel.load_weights('saved_models/weights.transfer.best.hdf5')

(IMPLEMENTATION) Test the Model

Try out your model on the test dataset of dog images. Ensure that your test accuracy is greater than 60%.

In [31]:
### TODO: Calculate classification accuracy on the test dataset.
# get index of predicted dog breed for each image in test set
t_predictions = [np.argmax(tmodel.predict(np.expand_dims(feature, axis=0))) for feature in test_transfer]

# report test accuracy
test_accuracy = 100*np.sum(np.array(t_predictions)==np.argmax(test_targets, axis=1))/len(t_predictions)
print('Test accuracy: %.4f%%' % test_accuracy)
Test accuracy: 85.0478%

(IMPLEMENTATION) Predict Dog Breed with the Model

Write a function that takes an image path as input and returns the dog breed (Affenpinscher, Afghan_hound, etc) that is predicted by your model.

Similar to the analogous function in Step 5, your function should have three steps:

  1. Extract the bottleneck features corresponding to the chosen CNN model.
  2. Supply the bottleneck features as input to the model to return the predicted vector. Note that the argmax of this prediction vector gives the index of the predicted dog breed.
  3. Use the dog_names array defined in Step 0 of this notebook to return the corresponding breed.

The functions to extract the bottleneck features can be found in extract_bottleneck_features.py, and they have been imported in an earlier code cell. To obtain the bottleneck features corresponding to your chosen CNN architecture, you need to use the function

extract_{network}

where {network}, in the above filename, should be one of VGG19, Resnet50, InceptionV3, or Xception.

In [32]:
### TODO: Write a function that takes a path to an image as input
### and returns the dog breed that is predicted by the model.
from extract_bottleneck_features import *

def Xception_predict_breed(img_path):
    # extract bottleneck features
    bottleneck_feature = extract_Xception(path_to_tensor(img_path))
    # obtain predicted vector
    predicted_vector = tmodel.predict(bottleneck_feature)
    # return dog breed that is predicted by the model
    return dog_names[np.argmax(predicted_vector)]

Step 6: Write your Algorithm

Write an algorithm that accepts a file path to an image and first determines whether the image contains a human, dog, or neither. Then,

  • if a dog is detected in the image, return the predicted breed.
  • if a human is detected in the image, return the resembling dog breed.
  • if neither is detected in the image, provide output that indicates an error.

You are welcome to write your own functions for detecting humans and dogs in images, but feel free to use the face_detector and dog_detector functions developed above. You are required to use your CNN from Step 5 to predict dog breed.

Some sample output for our algorithm is provided below, but feel free to design your own user experience!

Sample Human Output

(IMPLEMENTATION) Write your Algorithm

In [33]:
### TODO: Write your algorithm.
### Feel free to use as many code cells as needed.

files, targets = load_dataset('dogImages/train')

def display_image(path):
    if path is None:
        return
    
    img = cv2.imread(path)
   
    # convert BGR image to RGB for plotting
    cv_rgb = cv2.cvtColor(img, cv2.COLOR_BGR2RGB)

    # display the image, along with bounding box
    plt.imshow(cv_rgb)
    plt.show()

def display_breed(breed):
    if not breed:
        return
    display_image(next((x for x in files if breed in x), None))
    
def have_fun(image_path):
    print("Input image: %s" % i)
    display_image(image_path)
    is_human = face_detector(image_path)
    is_dog = dog_detector(image_path)

    if is_human:
        print("Looks like a homo sapiens")
        breed = Xception_predict_breed(image_path)
        print("... and you look like a %s" % breed)
        ### TODO: Write your algorithm.
### Feel free to use as many code cells as needed.

files, targets = load_dataset('dogImages/train')

def display_image(path):
    if path is None:
        return
    
    img = cv2.imread(path)
   
    # convert BGR image to RGB for plotting
    cv_rgb = cv2.cvtColor(img, cv2.COLOR_BGR2RGB)

    # display the image, along with bounding box
    plt.imshow(cv_rgb)
    plt.show()

def display_breed(breed):
    if not breed:
        return
    display_image(next((x for x in files if breed in x), None))
    
def have_fun(image_path):
    print("Input image: %s" % i)
    display_image(image_path)
    is_human = face_detector(image_path)
    is_dog = dog_detector(image_path)

    if is_human:
        print("Looks like a homo sapiens")
        breed = Xception_predict_breed(image_path)
        print("... and you look like a %s" % breed)
        display_breed(breed)
    elif is_dog:
        print("bark...bark...")
        breed = Xception_predict_breed(image_path)
        print("Predectited breed: %s" % breed)
        display_breed(breed)
    else:
        print("No human faces or dogs detected ¯\_(ツ)_/¯")

Step 7: Test Your Algorithm

In this section, you will take your new algorithm for a spin! What kind of dog does the algorithm think that you look like? If you have a dog, does it predict your dog's breed accurately? If you have a cat, does it mistakenly think that your cat is a dog?

(IMPLEMENTATION) Test Your Algorithm on Sample Images!

Test your algorithm at least six images on your computer. Feel free to use any images you like. Use at least two human and two dog images.

Question 6: Is the output better than you expected :) ? Or worse :( ? Provide at least three possible points of improvement for your algorithm.

Answer: Output meets my expectations. :) It clearly has entertaining and educational value. Possible points of improvement:

  • Improve human face detector (FN) to handle better images with obstructed human faces and human profiles.
  • Improve human face detector (FP) by eliminating misclassification of dog faces as human ones. Example: dogImages/test/082.Havanese/Havanese_05585.jpg is detected as a human.
  • Note that lfw/Zhang_Wenkang/Zhang_Wenkang_0001.jpg, lfw/Zurab_Tsereteli/Zurab_Tsereteli_0001.jpg and lfw/Claudia_Coslovich/Claudia_Coslovich_0001.jpg resemble the same dog breed Petit_basset_griffon_vendeen. I suspect that it is due to image backgroud influencing breed prediction. Looks like extracting just human face from an image and feeding it to Xception_predict_breed can help mitigate such side effect.
  • Improve model's stability as I noticed diferent predicted results are returned from independent trainings. Would consider How to Get Reproducible Results with Keras
In [35]:
# TODO: Execute your algorithm from Step 6 on
## at least 6 images on your computer.
## Feel free to use as many code cells as needed.
import random

images = ["lfw/Zumrati_Juma/Zumrati_Juma_0001.jpg",
          "lfw/Claudia_Coslovich/Claudia_Coslovich_0001.jpg",
          "lfw/Yossi_Beilin/Yossi_Beilin_0002.jpg",
          "dogImages/test/035.Boykin_spaniel/Boykin_spaniel_02497.jpg",
          "lfw/Zurab_Tsereteli/Zurab_Tsereteli_0001.jpg",
          "dogImages/test/029.Border_collie/Border_collie_02051.jpg",
          "lfw/Zhang_Wenkang/Zhang_Wenkang_0001.jpg",
          "dogImages/test/101.Maltese/Maltese_06766.jpg",
          "dogImages/test/082.Havanese/Havanese_05585.jpg",
          "lfw/Recep_Tayyip_Erdogan/Recep_Tayyip_Erdogan_0001.jpg",
          "python.jpg"]
for i in images:
    have_fun(i)
    print("="*60)
Input image: lfw/Zumrati_Juma/Zumrati_Juma_0001.jpg
Looks like a homo sapiens
... and you look like a Dachshund
============================================================
Input image: lfw/Claudia_Coslovich/Claudia_Coslovich_0001.jpg
Looks like a homo sapiens
... and you look like a Petit_basset_griffon_vendeen
============================================================
Input image: lfw/Yossi_Beilin/Yossi_Beilin_0002.jpg
Looks like a homo sapiens
... and you look like a Dachshund
============================================================
Input image: dogImages/test/035.Boykin_spaniel/Boykin_spaniel_02497.jpg
bark...bark...
Predectited breed: Boykin_spaniel
============================================================
Input image: lfw/Zurab_Tsereteli/Zurab_Tsereteli_0001.jpg
Looks like a homo sapiens
... and you look like a Petit_basset_griffon_vendeen
============================================================
Input image: dogImages/test/029.Border_collie/Border_collie_02051.jpg
bark...bark...
Predectited breed: Border_collie
============================================================
Input image: lfw/Zhang_Wenkang/Zhang_Wenkang_0001.jpg
Looks like a homo sapiens
... and you look like a Petit_basset_griffon_vendeen
============================================================
Input image: dogImages/test/101.Maltese/Maltese_06766.jpg
bark...bark...
Predectited breed: Maltese
============================================================
Input image: dogImages/test/082.Havanese/Havanese_05585.jpg
Looks like a homo sapiens
... and you look like a Bearded_collie
============================================================
Input image: lfw/Recep_Tayyip_Erdogan/Recep_Tayyip_Erdogan_0001.jpg
Looks like a homo sapiens
... and you look like a Petit_basset_griffon_vendeen
============================================================
Input image: python.jpg
No human faces or dogs detected ¯\_(ツ)_/¯
============================================================